🛰️ SentinelAI

🚨 AI-Powered Disaster Image Assessment Platform

Analyze disaster-related images using Google Gemini Vision, extract structured visual features, and use a Machine Learning severity classifier to estimate disaster severity.

Live Demo: http://3.110.196.251/

---

📖 Overview

SentinelAI is an end-to-end AI application designed to analyze disaster-related images and estimate their overall severity.

The system combines Vision AI, Machine Learning, REST APIs, image processing, containerization, and cloud deployment into a single application.

A user uploads an image through the Streamlit interface. The image is processed and sent to Google Gemini Vision, which analyzes the scene and identifies disaster-related indicators.

The resulting confidence values are converted into structured features and passed to a trained Random Forest classifier, which predicts the overall severity.

The application is deployed on AWS EC2 using Docker, Docker Compose, and Nginx.

---

🌐 Live Demo

🚀 Try SentinelAI:

http://3.110.196.251/

The live demo is hosted on an AWS EC2 instance.

---

✨ Features

📸 Image Analysis

• Upload disaster-related images
• Automatic image preprocessing
• RGB conversion
• Image resizing
• Unique image filenames
• Original and processed image storage

👁️ Vision AI

Uses Google Gemini Vision to analyze the uploaded image and identify:

• 🔥 Fire
• 💨 Smoke
• 🌊 Flood
• 🏢 Buildings
• 🛣️ Roads
• 🚗 Vehicles
• 👥 People
• 🌳 Trees
• 🏚️ Collapsed structures

Each detected feature includes a confidence score.

🤖 Machine Learning

The extracted vision features are passed to a trained Random Forest classifier to estimate:

• Disaster severity
• Prediction confidence

📊 Assessment

The application provides:

• Detected disaster indicators
• Feature confidence scores
• Overall severity
• Model confidence
• AI-generated scene summary
• Assessment report

⚙️ Backend API

FastAPI provides the backend REST API responsible for:

• Image processing
• Vision AI integration
• Feature extraction
• ML inference
• Assessment generation

🐳 Containerized Deployment

The application is separated into:

• Backend Docker container
• Frontend Docker container

Both services are managed using Docker Compose.

☁️ Cloud Deployment

SentinelAI is deployed on:

• AWS EC2
• Docker
• Docker Compose
• Nginx reverse proxy

---

🧠 System Architecture

                         User
                           │
                           ▼
                  Streamlit Dashboard
                           │
                           ▼
                    Image Upload
                           │
                           ▼
                  Image Preprocessing
                           │
                           ▼
                 FastAPI Backend
                           │
                           ▼
                  Gemini Vision API
                           │
                           ▼
             Disaster Feature Extraction
                           │
                           ▼
              Structured Feature Vector
                           │
                           ▼
                Random Forest Model
                           │
                           ▼
               Severity + Confidence
                           │
                           ▼
                 Assessment Results
                           │
                           ▼
                  Streamlit Dashboard